Kueue: A Kubernetes-native Job Queueing

on YouTube ↗35m

" summary

Kueue is a Kubernetes-native queuing controller that actually enforces job-level admission using the existing batch/v1 Job API: it introduces Queue (namespace pointer) and ClusterQueue (cluster-scoped quotas with flavors), forces jobs to start suspended, injects node selectors/tolerations and unsuspends admitted jobs, and implements borrowing cohorts and fair sharing. Version 0.1 is released and works for v1 Jobs; planned features include a CapacityRequest API for autoscaler integration and custom-workload support via a lightweight Workload abstraction.

AI-written summary · about

" chapters (35)

1. 0:00Introduction
2. 0:47Problem description
3. 1:12What is a job
4. 2:24Batch workloads
5. 3:25Flexible on multiple dimensions
6. 4:37The problem
7. 4:56Microbatch jobs
8. 6:08What is Job Queueing
9. 7:42What we want to solve
10. 8:50Why propose a new controller
11. 9:54The concept of a queue
12. 10:55Design principle
13. 13:02Job API
14. 13:57Resource Model
15. 15:22Multiple namespaces
16. 15:51Multiple cluster queues
17. 16:17Borrowing cohorts
18. 16:41Queue API
19. 17:17Cluster Queue API
20. 17:57Dakota
21. 20:31Borrowing Cohort
22. 21:14How it works
23. 21:39queuing controller
24. 22:11existing controller
25. 22:24handling custom workloads
26. 22:54workload API
27. 23:31workload controller
28. 24:41autoscaling
29. 25:55why
30. 27:22integration
31. 27:38capacity requests
32. 28:07pod management
33. 29:23release
34. 30:24questions
35. 32:04how does the queue interact with index jobs

" related talks

• 29mBuilding a Batch System for the Cloud with KueueKante Yin, Aldo Culquicondor · Kubernetes Batch + HPC Day Europe 2023
• 33mBatch Systems in Production with Kueue: Multi-Tenancy and FungibilityAldo Culquicondor, Yuki Iwai · Kubernetes AI + HPC Day North America 2023
• 28mMultitenancy and Fairness at Scale with Kueue: A Case StudyAldo Culquicondor, Rajat Phull · Cloud Native + Kubernetes AI Day NA 2024
• 24mAccelerate Your AI/ML Workloads With Topology-Aware Scheduling in KueueMichał Woźniak, Yuki Iwai · Cloud Native + Kubernetes AI Day Europe 2025
• 37mIntroduction to the Kubernetes WG BatchAldo Culquicondor, Abdullah Gharaibeh, Alex Wang · KubeCon + CloudNativeCon Europe 2022
• 05mKueue: Save Some QPS for the Rest of Us! How To Manage 100k Updates Pe...P. Bundyra · KubeCon + CloudNativeCon Europe 2025
• 34mWG-Batch Updates: What’s New and What Is Next?Marcin Wielgus · KubeCon + CloudNativeCon Europe 2025
• 30mSIG Scheduling Intro & UpdatesKensei Nakada, Maciej Skoczeń · KubeCon + CloudNativeCon Europe 2025